Multilingual data file Oracle - 8.1.7

Question

I am using 'sqlldr' to insert data into Oracle 8.1.7 database(charset UTF-8, OS=Win NT) from a datafile encoded in UTF-8. This datafile contains data in English, Japanese, Chinese languages. All rows were inserted into the database but I couldn't see Japanese and Chinese data while making a SELECT from the table.
I know I have to set some parameter called NLS_LANG but where should I set it and to what before starting 'sqlldr' utility?

Answer 1

I think your problem is that you are experiencing an unwanted conversion. Your nls_lang setting needs to be UTF8 from where you are running SQLLDR. I advise you to carefully look at the National Language Support Guide before proceeding. But to give you an answer. Use the following views to check your current environment.
NLS_DATABASE_PARAMETERS - Shows the current NLS parameters of the database, including the database character set.
NLS_SESSION_PARAMETERS - parameters set by NLS_LANG or alter session
From what you are telling me you are going utf8 file -> utf8 database character set
Before running sqlldr you need to change the nls_lang setting for your session to utf8 to avoid conversion.
The client character set is set by the NLS_LANG parameter. This has the format
<language>_<territory>.<client character set>
On the UNIX platform, NLS_LANG is set as an environment variable.
On the Windows platform, NLS_LANG is set globally in the registry and you can override it by setting the NLS_LANG local environment in your command prompt.
The NLS_LANG parameter has three components(language, territory, and charset) in the form:
NLS_LANG = language_territory.charset
The following lists some commonly used values for NLS_LANG.
JAPANESE_JAPAN.JA16SJIS
AMERICAN_AMERICA.UTF8
You must also set the characterset to utf8 in sqlldr. Syntax is: CHARACTERSET char_set_name
null